warisgill/TraceFL
TraceFL is a novel mechanism for Federated Learning that achieves interpretability by tracking neuron provenance. It identifies clients responsible for global model predictions, achieving 99% accuracy across diverse datasets (e.g., medical imaging) and neural networks (e.g., GPT).
TraceFL helps machine learning engineers and researchers understand which clients contribute most to a global model's predictions in a federated learning setup. It takes a trained global model and, for any given prediction, identifies the specific client (e.g., a hospital's dataset) that was most influential. This allows the practitioner to debug issues, assess client contributions, and enhance model reliability without ever seeing raw client data.
No commits in the last 6 months.
Use this if you are a machine learning engineer or researcher working with federated learning models and need to understand the source of a model's predictions for debugging, accountability, or quality control.
Not ideal if you are working with traditional centralized machine learning models, as its core value is in tracing contributions across distributed clients.
Stars
10
Forks
—
Language
Python
License
MIT
Category
Last pushed
Nov 12, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/warisgill/TraceFL"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
flwrlabs/flower
Flower: A Friendly Federated AI Framework
JonasGeiping/breaching
Breaching privacy in federated learning scenarios for vision and text
anupamkliv/FedERA
FedERA is a modular and fully customizable open-source FL framework, aiming to address these...
zama-ai/concrete-ml
Concrete ML: Privacy Preserving ML framework using Fully Homomorphic Encryption (FHE), built on...
p2pfl/p2pfl
P2PFL is a decentralized federated learning library that enables federated learning on...